OBJECTIVES: To assess the performance of circulating vascular endothelial growth factor (VEGF) levels as a tool for diagnosing giant cell arteritis (GCA) in a cohort of patients referred for assessment of suspected GCA. METHODS: We selected 298 patients recruited to the multicentre study Temporal Artery Biopsy versus Ultrasound in diagnosis of suspected GCA (TABUL). In a random subset of 26 biopsy-proven GCA cases and 26 controls, serum from weeks 0, 2 and 26 was analysed for VEGF concentration using ELISA. VEGF concentration at week 0 was used to generate a receiver-operating characteristic curve and thereby identify a cut-off for an abnormal result which was used to analyse the full patient cohort. Sections of paraffin-embedded temporal artery were stained by immunohistochemistry for VEGF. RESULTS: The mean (95% CI) VEGF concentration at week 0 was 873 pg/mL (631 to 1110) in 26 patients versus 476 pg/mL (328 to 625) in 26 controls (p=0.017). This difference was not observed at any other time point. The optimal cut-off of 713 pg/mL was applied to the whole patient cohort (n=298), yielding sensitivity of 32% and specificity of 85%. This was not improved by combination with any clinical parameters. When patients with biopsy-proven GCA were compared with controls, sensitivity was 58% and specificity remained 85%. Sections of biopsy from biopsy-positive GCA showed intense staining in the adventitia which was not seen in controls. CONCLUSIONS: Serum VEGF concentration predicts biopsy positivity but is not useful for differentiating clinical cases of GCA from controls. Further studies into VEGF as a prognostic marker and therapeutic target are warranted. TRIAL REGISTRATION NUMBER: NCT00974883; Post-results.